Post on 23-Jul-2020
transcript
Contents lists available at ScienceDirect
Composite Structures
journal homepage: www.elsevier.com/locate/compstruct
Dynamic response of flexible hybrid electronic material systems
Nicholas C. Searsa, John Daniel Berriganb, Philip R. Buskohlb, Ryan L. Harnea,⁎
a Department of Mechanical and Aerospace Engineering, The Ohio State University, Columbus, OH 43210, USAb Soft Materials Branch, Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright Patterson Air Force Base, OH 45433, USA
A B S T R A C T
Flexible hybrid electronic (FHE) material systems embody the intersection of compliant electrical networks and functional material architectures. For a wide varietyof future applications, FHE material systems will be subjected to dynamic mechanical stresses, such as for motion monitoring or for vibration isolation. Consequently,an understanding is required on how these new classes of material systems may respond mechanically and electrically when under states of high-cycle and high-frequency loads. Here, conductive silver microflake ink is interfaced with elastomeric geometries programmed with specific strain responses. Changes in electricalresistance under cyclic displacements are shown to depend on the heat generated by electrical current flow and on the thermal heat generation promoted by the pre-strain on the material system. Configurations subject to high static pre-strains and large strain rates exhibit greater increases in temperature and resistance, whereas anear constant conductivity is manifest in FHE material systems with compositions that reduce static local strains despite high engineering pre-strain application.These results may guide future efforts to understand the resistance change in conductive ink networks and expand the use of flexible hybrid electronic materialsystems into myriad dynamic application environments.
1. Introduction
Research into flexible hybrid electronics (FHEs) has flourished dueto the new opportunities made possible by the introduction of com-pliance into electronic components. FHEs have been implemented infields ranging from human health monitoring to soft robotics for theability to conduct electrical signals while undergoing large strain [1].FHEs can be composed of a network of conductive flakes or particlesembedded in a flexible material matrix [2–4], which together constituteconductive inks that serve as compliant, electrical conduits [4–6].Conductive ink-based FHEs have been formulated with a variety ofmetal flakes or particles, including copper [5], gold [6], and silver [4].Typical polymers for the material matrix of the inks include thermo-plastics due to desirable flexibility [4]. The dispersion of conductiveflakes or particles within a polymer matrix creates a flexible and con-ductive percolating network in which overall conductivity is de-termined by the proximity of contact between the microscale con-ductive constituents [7].
Despite the flexibility that FHEs provide, a disruption of the con-ductive network through mechanical or thermal stresses may result in areduction of electrical conductivity and thus reduction of operationaleffectiveness. In light of this, there has been an interest to characterizeand understand the mechanisms that contribute to changes of theconductivity of such networks. For example, large strains [8,9] andhigh strain rates [10,11] can disrupt the conductive network throughstress that may physically separate previously adjacent conducting
particles. Similarly, temperature changes within electrically conductivenetworks can result in thermal stresses due to the different thermalexpansion properties of the conductive constituents and the polymermatrix [12]. Large temperature fluctuations during fabrication [13] orthe heat generated by electric currents [12] have also demonstrated adisruption of current flow via the thermal stress. Thermal imaging hasrevealed direct evidence of how the current flow through electricallyconductive networks is influenced by the state of microstructural stressand strain [14].
In certain cases, the sensitivities of FHEs to stress and strain areexploited for functional applications that are unachievable with con-ventional electronics. For example, conductive inks may be leveraged todetect or measure strain through resistance changes that occur as aresult of a change in the strained conductive network [4,15–17]. In-terest in FHE 'wearables' has stimulated attention to methods of har-nessing FHEs to monitor muscle movements in speech [18] and heartrate [19]. Still, the full scope of opportunities for FHEs to build uponthe capabilities of conventional electronics has yet to be explored.
For instance, the influence of dynamics, such as high frequencycyclic forces, on the conductivity of conventional wired electronics iswell known [20,21]. This has inspired the use of tunable elastomericmaterial systems that leverage large deformation of microscopicstructural geometries to mitigate vibration energy transfer [22,23].With the elastomer matrix of conductive ink based FHEs, similar op-portunities may exist for FHEs to enhance the capabilities of conven-tional electronics by combining protective and electrically conductive
https://doi.org/10.1016/j.compstruct.2018.10.023Received 30 May 2018; Received in revised form 3 October 2018; Accepted 8 October 2018
⁎ Corresponding author.E-mail address: harne.3@osu.edu (R.L. Harne).
Composite Structures 208 (2019) 377–384
Available online 09 October 20180263-8223/ © 2018 Elsevier Ltd. All rights reserved.
T
functions. Therefore, the use of conductive ink-based FHEs withingeometries designed for vibration and shock energy mitigation presentsan ideal opportunity to create FHE material systems and investigatebehaviors manifest in dynamic excitation environments.
In order to achieve an understanding of how FHE material systemsmay respond under dynamic loads, there is a need to account for andmonitor transient behavior that may influence mechanical and elec-trical properties. Furthermore, it is important to uncover how transientbehavior may change with varying static strain distributions manifest incompressed FHE material systems [14]. For example, thermoplasticsare a common material of choice for the elastomer material matrix ofFHEs [24]. Yet the inherent viscoelasticity results in strain transfercharacteristics that depend upon the static and dynamic states of me-chanical stress [25]. Thermoplastics also exhibit softening under lowfrequency cyclic displacements due to the Mullin's effect [24] and thesoftening may be attributed to internal temperature rise from highfrequency mechanical loads [25–27]. Additionally, filler particleswithin thermoplastics contribute to the accumulation of internal heat[27], which may be significant when considering the effects of em-bedded metal flakes under strain in thermoplastics. Therefore, the cu-mulative influence of these behaviors on the electrical conductivity ofFHE networks in material systems subjected to high frequency dynamicloads is important to uncover.
Summarizing the state-of-the-art understanding of FHE propertiesadaptation resulting from mechanical loads, changes in static strain andstress distinctly tailor the microstructural characteristics of FHE net-works, which influence electrical conductivity. Yet, the nuanced me-chanisms that contribute to the precise electrical response under dy-namic mechanical loads are unknown. Therefore, this research seeks toreveal how the static strain characteristics of conductive networksembedded in material systems contribute to electrical conductivity andmechanical behaviors under dynamic stresses. Because a wide varietyof applications that may deploy FHEs involve relatively low frequencyharmonic stresses, such as at frequencies 50 Hz and less, this researchgives priority attention to this frequency regime.
This report is organized as follows. In Section 2, the FHE materialsystems fabricated and modeled here are described in detail. Then, inSection 3, a comprehensive investigation into the underlying physics ofelectrical resistance change within conductive inks undergoing dy-namic excitation is given. Finally, in Section 4, a summary of thefindings of this research and implications are presented.
2. Specimen descriptions and investigative methods
2.1. Specimen descriptions
In order to investigate the dynamic behaviors of FHE material sys-tems potentially suitable for vibration mitigation practices, two mate-rial system compositions are studied. Fig. 1(a) and (b) respectively
show the ‘strain-sensitive’ and ‘strain-insensitive’ material systems. Inthis work, the mechanical and electrical behaviors of the materialsystems are caused by uniaxial applied displacement from the verticaldirection according to the orientation of Fig. 1. Considering the loadingaxis, the strain-sensitive specimen of Fig. 1(a) is used to investigate thedynamic influences on conductive ink traces interfaced with geometriesthat maximize strain transfer to the ink trace by the snap-through re-sponse of the vertically-oriented buckling beam-like member [28]. Incontrast, the strain-insensitive specimen in Fig. 1(b) is used to in-vestigate the dynamic influences on conductive ink traces interfacedwith geometries that may limit significant strain transfer to the con-ductive ink trace through more gradual collapse of the geometry causedby the applied displacement [29].
The FHE material systems shown in Fig. 1(a) and (b) represent di-verse compositions that permit close study of electrical characteristicsof conductive inks subjected to static pre-strain and dynamic mechan-ical loads. For this research, an ink composed of silver (Ag) microflakesembedded in a thermoplastic polyurethane (TPU) matrix is prepared(Ag-TPU ink). The electrical resistance of the Ag-TPU ink is measuredthrough wire leads, as shown in Fig. 1(a) and (b). The specimen geo-metries are composed of a commercial 3D printed, thermoplasticpolymer (Stratasys material Tangoblackplus, FLX980). For more in-formation regarding the preparation and fabrication of the specimens,see Supporting Information Section 1.
2.2. Experimental methods and FE model description
The dynamic properties of the FHE material systems shown in Fig. 1are evaluated at engineering pre-strains ε, defined as the applied dis-placement over initial specimen height. The three values of pre-strainused to examine the FHE material systems of this work are selected tocharacterize the buckling behavior of the strain-sensitive specimen: pre-buckled, near the buckling point, and post-buckled. These configura-tions correspond to engineering pre-strains of ε =4%, 8%, and 14%,respectively. These pre-strain magnitudes are used to investigate thestrain-sensitive and strain-insensitive specimens, respectively shown inFig. 2(a) and (b).
At each of the static pre-strain configurations shown in Fig. 2(a) and(b), the specimens are subjected to sinusoidal displacements applied byan electrodynamic shaker (LDS V408) that is controlled by a controller(Vibration Research VR9500) and amplifier (Crown XLS1500). As thespecimen undergoes harmonic displacement input, the input and outputdynamic forces are measured by a matched pair of force transducers(PCB 208C01) that start and terminate the loading fixture. The con-ductive ink trace resistance is simultaneously monitored with a voltagedivider circuit. The components of the experimental setup are labeled inFig. 2(a). This experimental setup is employed to characterize how themechanical and electrical properties are influenced by the pre-strainand applied dynamic displacements at frequencies of 50 Hz and less.
Fig. 1. (a) Strain-sensitive and (b) strain-insensitive specimens with conductive ink traces and wire leads to measure resistance.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
378
This represents a frequency range indicative of potential applications ofFHE materials systems as wearables, self-sensing elastomers andmounts, motion sensors, among other applied contexts. Resonance inthe strain-sensitive and strain-insensitive specimens occurs at fre-quencies in the 1000 s of Hz, where viscoelastic damping dominates andsuppresses unique dynamic behaviors. Consequently, the low fre-quency, sub-resonant regime with dynamic stresses of 50 Hz and less isconsidered in this investigation as the loading environment for thematerial systems.
In order to model the mechanical behaviors of the specimens ob-served in experiments, two-dimensional (2D) plane strain, finite ele-ment (FE) models of the specimens are generated in ABAQUS. A Neo-Hookean, hyperelastic material model serves as the basis for the FEmodel by providing the necessary material properties of the specimens.The hyperelastic material model uses the following properties that areempirically identified: density ρ =1112 kg.m−3, Poisson’s ratioυ =0.49, shear modulus μ =248 kPa, and bulk modulus κ =16.6MPa.Additionally, in order to model the frequency and time dependent in-fluences observed experimentally, a Prony series, viscoelastic materialmodel is utilized. Details about the Prony series coefficient identifica-tion procedure is given in the Supporting Information Section 2. For theFE modeling of the dynamic loading experiments, a two step procedureis employed. First, a dynamic-implicit model formulation applies static
pre-strain to specimens, while a second dynamic-explicit study stepapplies the harmonic displacement input. In order to replicate experi-ments, one side of each specimen is fixed (output side) while pre-strainsand harmonic displacements are applied to the opposite end (inputside). Self-contact by tangential friction penalty coefficient of 90% isapplied to all edges of the 2D FE model domain. Prior to obtaining finalresults for processing, a mesh convergence study with CPS4R elementsis undertaken to ensure that the FE model outcomes are consistent inquantitative values by a suitably refined mesh.
As shown in Fig. 2(a) and (b), the static deformation of the specimenconfigurations in experiments is accurately reconstructed by the FEmodel. In the FE model results of Fig. 2(a,b), the shading is coloredaccording to maximum principal strain. The agreement between thestatic deformations observed experimentally and computationally forthe strain-sensitive and -insensitive specimens in Fig. 2(a) and (b), re-spectively, indicates that the static strain distributions identified by theFE models may likewise be accurate representations.
In Fig. 2(c), for a 0.25mm peak-to-peak applied harmonic dis-placement, average experimental data is shown according to datamarkers with range bars indicating maximum and minimum measure-ments across three separate experiments. Experimental results show anincrease in output force with frequency of the applied displacement.Furthermore, as shown in Fig. 2(c), for a 0.25mm peak-to-peak applied
Fig. 2. (a) Strain-sensitive specimen, with experimental setup indicated, and (b) strain-insensitive specimen with FE model static pre-strain deformation shapes,colored according to maximum principal strain at engineering pre-strains ε =4%, 8%, and 14%. (c) Comparison of average output force for 0.25 mm peak-to-peakexcitation applied at frequencies between 3 and 50 Hz for experiments, a hyperelastic model, and a hyperelastic and viscoelastic model. Maximum and minimummeasurements are indicated by range bars.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
379
harmonic displacement, the hyperelastic and viscoelastic materialmodel shows increasing average output force under dynamic excitation,compared to a hyperelastic model that is frequency independent. Thecombined hyperelastic and viscoelastic material model exhibits astronger correlation to experimental data (R2 =0.77) than the hyper-elastic material model alone (R2 =0.0087). Thus, the addition of theviscoelastic material model is essential to the exploration of the stress,strain, and deformation dynamics of these specimens and matches wellwith experiments. The combined hyperelastic and viscoelastic materialmodel is used in the ensuing FE investigations that augment and elu-cidate experimental findings.
3. Results and discussion
3.1. High-cycle and high-frequency loading influences on FHE materialsystem behaviors
In order to evaluate the FHE material systems under high frequencydynamic mechanical loads, specimens undergo cyclic displacementexperiments. For these experiments, a 0.25mm peak-to-peak displace-ment is applied at a 25 Hz excitation frequency to specimens under thepre-strains identified in Fig. 1(a) and (b) for 10,000 displacement cy-cles. Before dynamic excitation is applied, specimens are allowed tocome to electrical equilibrium until the pre-strained specimen re-sistance measurements are approximately constant, which typicallyoccurs within 10min under static stress conditions. At this time, initialelectrical resistance measurements corresponding to each configuration
are recorded and reported as R0.As shown in Fig. 3(a) and (b), the force transfer function computed
as the ratio of dynamic output force to dynamic input force, increasesby less than 0.8% for all specimens and configurations by displacementcycle 10,000. On the other hand, the trends in resistance change inFig. 3(c) and (d) differ among the specimen geometries and pre-strainconfigurations. As shown in Fig. 3(c), the resistances of the conductiveink trace in ε =4% and 8% configurations of the strain-sensitive spe-cimen show a 35% and 8% increase in resistance by displacement cycle100, respectively, followed by gradual increases in resistance over theduration of the experiments. Yet, the resistance of the conductive inktrace in the ε =14% configuration does not indicate sudden resistanceincrease after 100 displacement cycles, and in fact gradually decreasesin resistance value over the remaining duration of the experiment. Forthe strain-insensitive specimen, the resistances of the conductive inktraces remain nearly constant over the duration of the experiment forall configurations, as shown in Fig. 3(d).
Previous studies have revealed that conductive networks may bedisrupted by the application of an instantaneous stress, which results inan increase in initial resistance [30,31]. The application of an in-stantaneous stress is generated at the beginning of each experimentwith the onset of harmonic displacement. Such stress may explain thelarge increases in resistance associated with the strain-sensitive spe-cimen for pre-strain conditions ε =4% and ε =8%. This is explored forthe specimens in this research in greater detail with the use of FE si-mulations in Supporting Information Section 2 to help elucidate thebehaviors observed in Fig. 3(c,d). It is found that greater peak stress
Fig. 3. Force transfer function is shown for ε =4%, 8%, and 14% configurations of the (a) strain-sensitive and (b) strain-insensitive specimens for a 10,000 cycledisplacement experiment with 0.25mm peak-to-peak applied displacement. Normalized resistance change is shown for ε =4%, 8%, and 14% configurations of the(c) strain-sensitive and (d) strain-insensitive specimens.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
380
changes within the conductive ink traces are associated with greaterinitial resistance changes, which may explain the trends in Fig. 3(c,d)for the respective relative resistance change for the FHE material sys-tems. On the other hand, a greater number of influences may still col-lectively yield such behaviors seen in Fig. 3(c,d), so that future work isdesired.
Resistance changes in conductive networks under cyclic mechanicalloading have previously been characterized with respect to low fre-quency excitations, wherein the findings reveal increases in electricalresistance due to accumulated disruptions of the conductive network[8,32]. Therefore, a need remains to explain the resistance trends ofFig. 3(c,d) that are shown to increase, decrease, or remain constant asthe number of high frequency displacement cycles increases. To thisend, an investigation is needed to study the influences of static pre-strain on the electrical resistance during cyclic displacement experi-ments.
3.2. Heat generation influence on resistance change
Mechanical strains can cause electrical resistance changes in FHEsthrough disruptions of the conductive network [4,15–17]. For instance,supporting experiments undertaken here exemplify the visible de-struction to the conductive ink network before and after the applicationof a static pre-strain, as shown in Supporting Information Section 3. Yet,research into the failure mechanisms of silver microflake-based elec-trical connections under thermal and mechanical influences shows thatthe efficacy of the conductive network is also strongly influenced bytemperature [33]. Although temperature is not usually considered incyclic deformation experiments of FHEs, the use of thermoplastics inthe FHE material systems studied here may cultivate relevant thermo-elastic phenomena under high frequency mechanical loads [27], whileheat generated due to electric currents passing through the conductiveink traces cannot be neglected [12]. In order to explore the influence oftemperature changes within the conductive ink traces, average ink tracetemperatures are measured with a thermal imaging camera (FLIR C2)during cyclic displacement experiments. Before each experiment, spe-cimens are allowed to come to thermal and electrical equilibrium. Atthis time, average ink trace temperatures and resistances are measuredto record baseline, constant values. Thermal images are shaded withcoloring according to temperature changes from −2 to +7 degrees C,with respect to the baseline temperature distributions.
Thermal images of the specimens are shown for displacement cycles0 and 10,000 for each pre-strain configuration, Fig. 4(a,c). As shown inFig. 4(a) for the strain-sensitive specimen, heat is noticeable within theconductive ink traces before and after experiments due to passing small,constant electrical current used for the resistance measurement in thevoltage divider circuit. The largest temperature changes of the strain-sensitive specimen appear to be concentrated within the conductive inktraces as well. For instance, the conductive ink traces in ε =4% andε =8% configurations of the strain-sensitive specimen show significantheat generation in Fig. 4(a), indicated by changes in color according tothe temperature change scale. Furthermore, as is most clearly ex-emplified by the ε =4% strain-sensitive specimen, the input side of thespecimen that is dynamically displaced by the input, corresponding tothe bottom of the specimen as shown in Fig. 4(a), experiences a largerincrease in temperature than that of the output side. Larger increases intemperature for the input sides of the specimens compared to theoutput sides may explain the increases in transfer function seen inFig. 3(a) and (b) due to a softening effect of elastomers that occurs withinternal heat generation [25,26]. In other words, a softening of outputforce with a greater softening of input force results in an increase inoutput-to-input transfer function over the duration of the experiment.In fact, this hypothesis is supported by complementary data in theSupporting Information Section 4 wherein the FHE material systemswithout conductive traces are also dynamically displaced at 25 Hz for10,000 displacement cycles and likewise show steady transfer function
increase with increase in displacement cycles. Yet, as shown in Fig. 3(a)and (b), these changes are relatively minor and result in a less than0.8% change in transfer function for both specimens.
Although the force transfer functions of the experimental specimensmay change only slightly due to internal heat generation, the thermo-elastic heat generation properties of the conductive ink may influenceelectrical resistance in a more significant way. For instance, polymerswith added fillers such as silver have been shown to possess greaterthermal conductivity than those of the unfilled polymer. In this way,the conductive ink traces may conduct more heat than the elastomer ofthe specimen geometry [34]. It has been shown that large temperaturescan disrupt conductive networks through thermal fatigue cracking andaccumulated strain [33]. This results in a degradation or failure of theconductive networks [12,13]. The thermoelastic heat generation inconductive ink traces is thought to be due to the thermal stresses as-sociated with two materials of different thermal expansion coefficients[12]. Thus, it may be concluded that temperature changes within theconductive ink traces, composed of metal microflakes within a ther-moplastic material matrix, result in the resistance changes measuredduring cyclic displacement experiments.
Indeed, as shown in Fig. 4(b), changes in average ink trace tem-perature of +5.3 and +2.8 degrees C are measured for the ε =4% andε =8% configurations of the strain-sensitive specimen, respectively. Onthe other hand, a change in average ink trace temperature of −0.5degrees C is measured for the ε =14% strain-sensitive specimen. Thesethermal and electrical behaviors contrast greatly with the results ob-tained for the strain-insensitive specimen. Fig. 4(c) shows thermalimages at displacement cycles 0 and 10,000 for the strain-insensitivespecimen. Heat is concentrated within the conductive ink traces beforeand after experiments, mostly associated with the electric current flow.Yet, unlike the ε =4% and 8% configurations of the strain-sensitivespecimen, all configurations of the strain-insensitive specimen showlittle to no heat generation within the conductive ink trace by dis-placement cycle 10,000. Measurements of the average conductive inktrace temperature of the strain-insensitive specimen in Fig. 4(d) showtemperature changes of +0.5, +0.6, and −0.9 degrees C for ε =4%,8%, and 14% configurations, respectively.
The thermal images in Fig. 4(a,c) and average temperature changesof Fig. 4(b,d) suggest that the extent to which thermal stress is gener-ated within the ink traces may be determined in part by specimenconfiguration that corresponds to the local, static strain in the con-ductive ink trace. For instance, as shown in the FE model deformationshapes of the strain-sensitive specimen in Fig. 2(a), the beam ontowhich the conductive ink trace is applied exhibits decreasing localprincipal strain with increasing compressive pre-strain for ε =4%, 8%,and 14% configurations. The state of local strain for ε =4% and 8%configurations, under dynamic mechanical loads, generates heatthrough thermal stresses that disrupt the conductive network so as tocontribute to the resistance changes at displacement cycle 10,000 of72% and 20%, respectively, as shown in Fig. 3(c). On the other hand,the FE model deformation shapes of the strain-insensitive specimen inFig. 1(b) show that all configurations of the strain-insensitive specimenexhibit smaller local, static principal strains in the conductive ink tracebeam. Therefore, the strain-insensitive specimen experiences lessthermal stress and exhibits average conductive ink trace temperaturechanges of less than 1 degree C in magnitude, as seen in Fig. 4(c,d). As aresult, the strain-insensitive specimen shows negligible change in re-sistance over the duration of the experiment for all pre-strain config-urations, Fig. 3(d).
Yet, temperature changes of less than 1 degree C, as measured forthe strain-insensitive specimen in Fig. 4(d), are also measured for theε =14% strain-sensitive specimen, as shown in Fig. 4(b), and resistancedecreases by 4% by displacement cycle 10,000, as shown in Fig. 3(c).This suggests that the resistance change of the ε =14% strain-sensitivespecimen may not be due to thermal stress. The decrease in resistancemay be due to the increased conductivity associated with conductive
N.C. Sears et al. Composite Structures 208 (2019) 377–384
381
networks under high strain. For instance, Busfield et al. [35] and Ya-maguchi et al. [36] show that for conductive networks under quasi-static tensile strain, an increase in resistance, due to a disruption of theconductive network, can be followed by a decrease in resistance due toa rotational alignment of conductive particles induced by large strains.The highly deformed conductive ink trace beam in the ε =14% con-figuration of the strain-sensitive specimen, as shown in Fig. 2(a), sug-gests that such unique realignment of conductive particles may becultivated through highly deformed shapes and contribute to similarelectrical behavior. Therefore, in the highly deformed configuration ofthe ε =14% strain-sensitive specimen, the state of static pre-straincontributes to a decrease in electrical resistance with cyclic deforma-tion, as shown in Fig. 3(c).
The many distinguishing resistance and temperature changes amongspecimens and geometric configurations highlights the intricate influ-ences that static pre-strains have on the electrical behavior of the FHEmaterial systems. Under dynamic mechanical loads, conductive inktraces in configurations of high static strain generate internal heatthrough thermal stresses. Thermal stresses, due to the viscoelasticity of
the elastomer matrix and due to the thermal expansion mismatch be-tween the elastomer matrix and embedded metal flakes, result in dis-ruptions of the conductive network and increases in electrical re-sistance. On the other hand, conductive ink traces in configurations oflow internal static pre-strain under dynamic mechanical excitation donot generate significant heat through thermal stresses and show negli-gible resistance change. It is also possible that conductive ink traces inhighly deformed configurations, such as the strain-sensitive specimensubjected to ε =14% pre-strain, may show decreasing electrical re-sistance with cyclic displacement, due to an induced re-alignment of theconductive network rather than thermal influences. Future researchshould investigate the movement of metal flakes within conductive inktraces in highly deformed configurations under dynamic excitationsince these configurations may provide a unique combination of highstatic pre-strain and low thermal stress characteristics.
3.3. Frequency dependence on resistance change
In Section 3.2, the influence of dynamic mechanical excitation on
Fig. 4. Specimen architecture and deformation mode dictates intensity of thermal load. (a) and (c) Thermal images at displacement cycle number 0 and 10,000 forthe strain-sensitive and strain-insensitive specimens, respectively. (b) and (d) Change in average temperature of conductive ink trace for all configurations of thestrain-sensitive and strain-insensitive specimens, respectively.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
382
the resistance changes of conductive ink traces in FHE material systemsunder pre-strains of ε =4%, 8%, and 14% is investigated for a 25 Hzfrequency of applied displacement cycles. In order to understand howFHEs respond under high frequency dynamic loads, it is important toconsider the effects that the loading frequency may have on the elec-trical resistance behavior. This is evident when considering the influ-ence of displacement excitation frequency on the average output forceshown in Fig. 2(d). In order to investigate the influence of loadingfrequency on FHE material systems, the resistances of the specimen inktraces are measured over 20 s time intervals at frequencies ranging from3 to 50 Hz and with a 0.25mm peak-to-peak amplitude of applieddisplacement. Measurements are taken 2min after excitation begins toallow the specimen to come to a pseudo-steady state of resistancechange. Additionally, specimens are allowed to come to electricalequilibrium where resistance measurements are constant before eachexperiment. These equilibrated resistances are denoted as R0. The datamarkers in Fig. 5 correspond to the average resistance measured duringthe 20 s time interval, while the range bars indicate the maximum andminimum resistance measurements taken during that time period.
As shown in Fig. 5(a) for the strain-sensitive specimen, changes inelectrical resistance of +112%, +17%, and -4% are measured at a 3 Hzdisplacement excitation frequency for the ε =4%, 8%, and 14% con-figurations, respectively. The differences in electrical resistance mea-surements between specimen configurations at the same displacementexcitation frequency are in part due to the influence of instantaneousstress applied at the beginning of harmonic displacement, which de-pend on the static strain characteristics of each configuration. This isdiscussed in Section 3.1 of the main text and Section 2 of the SupportingInformation.
Additionally, as shown in Fig. 5(a) for the strain-sensitive specimen,resistance measurements increase between 3 and 50 Hz by +40%,+8%, and +4% for the ε =4%, 8%, and 14% strain-sensitive specimenconfigurations, respectively. The trend of increasing resistance withfrequency may be due to the frequency dependent properties of the FHEmaterial systems. On the other hand, the amount of resistance changebetween displacement excitation frequencies for the same configurationdepends on the static pre-strain characteristics. For instance, an in-crease in average output force according to displacement excitationfrequency, as shown in Fig. 2(d), is due to the viscoelastic materialproperties of the specimens. Increases in average output force withdisplacement excitation frequency suggest that the state of stress andstrain within the strain-sensitive specimen increase with frequency aswell. To explore this further, FE model simulations are implemented toinvestigate how the strain rate that the ink trace experiences dependson the configuration and displacement excitation frequency, see
Supporting Information Section 5. The frequency dependence of in-ternal strain within the strain-sensitive specimen becomes manifest asincreases in the state of strain within the conductive networks. Suchgrowth of local strain acts to disrupt the conductive network and in-creases electrical resistance with increasing displacement excitationfrequency. Furthermore, the influence of static strain characteristics canbe seen in comparisons of the electrical resistance behavior betweenstrain-sensitive specimen configurations. For example, the ε =4%strain-sensitive configuration is in a state of greater maximum principalstrain compared to the ε =8% and ε =14% configurations, seeFig. 2(a), which results in a greater change in resistance between dis-placement excitation frequencies.
It is important to note that the testing procedure may contribute tothe frequency dependent resistance response of the strain-sensitivespecimen in Fig. 5(a). After 2min of applied harmonic displacement, alarger increase in resistance may be expected for a 50 Hz displacementexcitation compared to a 3 Hz displacement excitation based on thenumber of displacement cycles that the conductive ink trace undergoes.Yet, if the number of displacement cycles determined the frequencydependent electrical resistance behavior in Fig. 5(a), the resistancemeasured for the ε =14% configuration of the strain-sensitive spe-cimen would be expected to decrease between 3 and 50 Hz, accordingto the results shown in Fig. 3(c). On the contrary, the resistance mea-surements of the ε =14% strain-sensitive specimen increase by 4%between 3 and 50 Hz. This suggests that the frequency dependent re-sistance behavior of the strain-sensitive specimen in Fig. 5(a) is pri-marily due to the frequency dependent strain characteristics of thespecimens, rather than the number of displacement cycles.
For the strain-insensitive specimen, the resistance measurements inFig. 5(b) remain constant within 2% resistance change for all dis-placement excitation frequencies and specimen pre-strain configura-tions. The low static principal strains in the conductive ink trace beamassociated with the strain-insensitive geometry, as shown in the FEmodel deformation shapes of Fig. 2(b), correspond to a similarly lowdynamic strain influence on the conductive ink network. As a result,resistance measurements remain nearly constant, regardless of thefrequency of the applied displacement and number of cycles. The re-sults presented here further demonstrate how the static pre-strainscontribute to the measured resistance change. With high static pre-strain, the strain-sensitive specimen configurations show frequencydependence through resistance changes of up to 40% between 3 and50 Hz displacement excitation frequencies. With lower static pre-strains, the strain-insensitive specimen maintains resistance changeswithin 2% of the initial values for all configurations and displacementexcitation frequencies considered.
Fig. 5. Resistance change for specimen configurations evaluated at frequencies ranging from 3 to 50 Hz for (a) strain-sensitive and (b) strain-insensitive specimens.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
383
4. Conclusions
Investigations into the dynamic properties of FHE material systemshighlight the need to account for transient mechanical, electrical, andthermal influences manifest in material and geometry selections. Cyclicdisplacement experiments suggest that thermal heat generation withinthe conductive ink traces contributes to conductive network disruptionand results in gradual resistance changes. In addition to the electriccurrent flow, heat generation arises from thermal stresses in the ink thatdepend on differences in thermal expansion properties of the elastomermatrix and the silver microflakes suspended therein. The amount ofheat generation is exacerbated by the pre-strain of the ink trace con-figuration. On the other hand, ink traces not subjected to high staticpre-strains or high dynamic strain rates may exhibit an electricallyrobust conductivity that does not appear to degrade with increasingcycles of dynamic load. Finally, it is revealed that the magnitude ofcumulative fatigue of the conductive ink traces is determined by thefrequency of excitation due to the frequency dependent materialproperties and configuration dependent strain rates. As a result, onemay envision exploiting a strain-sensitive FHE material system for re-sistance-based monitoring of dynamic response and cyclic stresses.Conversely, a nearly dynamic-independent conductivity may be culti-vated by strain-insensitive FHE that delivers electrically robust prop-erties even when subjected to high-cycle and high-frequency dynamicloads. These findings may guide attention to new concepts for FHEmaterial systems exploited in applications involving periodic stress,such as wearables for motion monitoring and vibration isolation prac-tices.
Acknowledgments
The authors acknowledge James Hardin of UES, Inc. and JimDeneault of UTC for discussions and advice pertaining to specimenfabrication. This work is supported by the Air Force ResearchLaboratory and Dayton Area Graduate/Faculty Fellowship Program viagrant number RX1-OSU-17-4.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.compstruct.2018.10.023.
References
[1] Bansal AK, Hou S, Kulyk O, Bowman EM, Samuel IDW. Wearable organic optoe-lectronic sensors for medicine. Adv Mater 2015;27:7638–44.
[2] Jiang H, Moon K, Li Y, Wong CP. Surface functionalized silver nanoparticles forultrahigh conductive polymer composites. Chem Mater 2006;18:2969–73.
[3] Zhang R, Moon K, Lin W, Wong CP. Preparation of highly conductive polymernanocomposites by low temperature sintering of silver nanoparticles. J Mater Chem2010;20(10):2018–23.
[4] Valentine AD, Busbee TA, Boley JW, Raney JR, Chortos A, Kotikian A, et al. Hybrid3D printing of soft electronics. Adv Mater 2017;29(40):1–8.
[5] Jeong S, Woo K, Lim S, Kim JS, Shin H, Xia Y, et al. Controlling the thickness of thesurface oxide layer on Cu nanoparticles for the fabrication of conductive structuresby ink-jet printing. Adv Funct Mater 2008;18:679–86.
[6] Chandra P, Singh J, Singh A, Srivastava A, Goyal RN, Shim YB. Gold nanoparticlesand nanocomposites in clinical diagnostics using electrochemical methods. JNanoparticles 2013;2013:1–12.
[7] Stauffer D, Aharony A. Introduction to percolation theory. 2nd ed. Philadelphia,
USA: Taylor & Francis; 1994.[8] Jahanshahi A, Salvo P, Vanfleteren J. Reliable strechable gold interconnects in
biocompatible elastomers. J Polym Sci 2012;50:773–6.[9] Wagner S, Lacour SP, Jones J, Hsu PI, Sturm JC, Li T, et al. Electronic skin: archi-
tecture and components. Physica E 2004;25:326–34.[10] Borghetti M, Serpelloni M, Sardini E, Pandini S. Mechanical behavior of strain based
on PEDOT:PSS and silver nanoparticles inks deposited on polymer subtrate by in-kjet printing. Sens Actuators, A 2016;243:71–80.
[11] Das NC, Chaki TK, Khastgir D. Effect of axial stretching on electrical resistivity ofshort carbon fibre and carbon black filled conductive rubber composites. Polym Int2002;51:156–63.
[12] Roberson DA, Wicker RB, MacDonald E. Microstructural characterization of elec-trically failed conductive traces printed from Ag nanoparticle inks. Mater Lett2012;76:51–4.
[13] Shemelya C, et al. Anisotropy of thermal conductivity in 3D printed polymer matrixcomposites for space based cube satellites. Addit Manuf 2017;16:186–96.
[14] Lessing J, Morin SA, Keplinger C, Tayi AS, Whitesides GM. Stretchable conductivecomposites based on metal wools for use as electrical vias in soft devices. Adv FunctMater 2015;25:1418–25.
[15] Khan S, Lorenzelli L. Recent advances of conductive nanocomposites in printed andflexible electronics. Smart Mater Struct 2017;26:1–23.
[16] Merilampi S, Bjorninen T, Haukka V, Ruuskanen P, Ukkonen L, Sydanheimo L.Analysis of electrically conductive silver ink on stretchable substrates under tensileload. Microelectron Reliab 2010;50:2001–11.
[17] Merilampi S, Bjorninen T, Ukkonen L, Ruuskanen P, Sydanheimo L. Embeddedwireless strain sensors based on printed RFID tag. Sensor Review 2011;31:32–40.
[18] Wang X, Gu Y, Xiong Z, Cui Z, Zhang T. Silk-molded flexible, ultrasensitive, andhighly stable electronic skin for monitoring human physiological signals. Adv Mater2014;26:1336–42.
[19] Khan Y, et al. Flexible hybrid electronics: direct interfacing of soft and hard elec-tronics for wearable health monitoring. Adv Funct Mater 2016;26:8764–75.
[20] Lau J, Schneider E, Baker T. Shock and vibration of solder bumped flip chip onorganic coated copper boards. J Electron Packag 1995;118:101–4.
[21] Zhou Y, Al-Bassyiouni M, Dasgupta A. Vibration durability assessment ofSn3.0Ag0.5Cu and Sn37Pb solders under harmonic excitation. J Electron Packag2009;131:011016.
[22] Wang P, Casadei F, Shan S, Weaver JC, Bertoldi K. Harnessing buckling to designtunable locally resonant acoustic metamaterials. Phys Rev Lett 2014;113:014301.
[23] Bishop J, Dai Q, Song Y, Harne RL. Resilience to impact by extreme energy ab-sorption in lightweight material inclusions constrained near a critical point. AdvEng Mater 2016;18:1871–6.
[24] Qi HJ, Boyce MC. Stress-strain behavior of thermoplastic polyurethane. Mech Mater2004;37:817–39.
[25] Mulliken AD, Boyce MC. Mechanics of the rate-dependent elastic-plastic deforma-tion of glassy polymers from low to high strain rates. Int J Solids Struct2005;43:1331–56.
[26] Rittel D. On the conversion of plastic work to heat during high strain rate de-formation of glassy polymers. Mech Mater 1999;31:131–9.
[27] Kultural SE, Eryurek IB. Fatigue behavior of calcium carbonate filled polypropyleneunder high frequency loading. Mater Des 2006;28:816–23.
[28] Harne RL, Wang KW. Harnessing bistable structural dynamics for vibration control,energy harvesting, and sensing. Chichester, UK: John Wiley & Sons Ltd; 2017.
[29] Mousanezhad D, Babaee S, Ebrahimi H, Ghosh R, Hamouda AS, Bertoldi K, et al.Hierarchical honeycomb auxetic metamaterials. Sci Rep 2015;5:1–8.
[30] Muth JT, Vogt DM, Truby RL, Menguc Y, Kolesky DB, Wood RJ, et al. Embedded 3Dprinting of strain sensors within highly stretchable elastomers. Adv Mater2014;26:6307–12.
[31] Wang L, Ding T, Wang P. Effects of instantaneous compression pressure on electricalresistance of carbon black filled silicone rubber composite during compressivestress relaxation. Compos Sci Technol 2008;68:3448–50.
[32] Graz I, Cotton DPJ, Lacour SP. Extended cyclic uniaxial loading of stretchable goldthin films on elastomeric substrates. Appl Phys Lett 2009;94:071902.
[33] Herboth T, Guenther M, Fix A, Wilde J. Failure mechanisms of sintered silver in-terconnections for power electronic applications. Electronic Components andTechnology Conference. 2013. p. 1621–7.
[34] Maiti SN, Ghosh K. Thermal characteristics of silver powder-filled polypropylenecomposites. J Appl Polym Sci 1994;52:1091–103.
[35] Busfield JJC, Thomas AG, Yamaguchi K. Electrical and mechanical behavior offilled rubber. III. dynamic loading and the rate of recovery. J Polym Sci2005;43:1649–61.
[36] Yamaguchi K, Busfield JJC, Thomas AG. Electrical and mechanical behavior offilled elastomers. I. The effect of strain. Journal of Polymer. Physics2003;41:2079–89.
N.C. Sears et al. Composite Structures 208 (2019) 377–384
384
1
Supporting Information
Title: Dynamic response of flexible hybrid electronic material systems
Nicholas C. Sears(1), John Daniel Berrigan(2), Philip R. Buskohl(2), and Ryan L. Harne(1)*
(1) Department of Mechanical and Aerospace Engineering, The Ohio State University, Columbus, OH 43210, USA
(2) Soft Materials Branch, Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright
Patterson Air Force Base, OH 45433, USA
* Corresponding author, email: harne.3@osu.edu
1 Specimen fabrication
The specimen substrates are created in a 3D printer (Stratasys Objet260 Connex3) using a Stratasys PolyJet
material, Tangoblackplus (FLX980). The conductive ink used for experimental specimens is composed of
silver (Ag) microflakes (Inframat Advanced Materials, 47MR-10F) and a thermoplastic polyurethane
(TPU) (BASF, 50126177). The Ag microflakes and TPU elastomer are combined and dissolved in an N-
Methyl-2-pyrrolidone (NMP) solvent (Alfa Aesar, A12260). The conductive ink solution is sonicated for
an hour at 60 rpm and centrifuged for 2 minutes at 2000 rpm before allowing the NMP to dissolve the TPU
over the course of two days. The conductive ink solution is centrifuged for 2 minutes at 2000 rpm before
every subsequent use. Once the combined mixture is applied to the specimens, the NMP evaporates and
Ag-TPU ink remains. The Ag-TPU ink formulation used includes Ag microflakes that constitute 35% of
the conductive ink volume (v%). Thus, the ink used in this research is 35v% Ag-TPU ink.
As shown in Figure S1(a), 34 gauge wire leads are connected to the ink trace on the surface of the
experimental specimens via internal channels 1.4 mm in diameter. Channels for the wire leads are printed
into the specimens and provide a stable connection for wire lead resistance measurements during dynamic
excitation. Ag-TPU ink is applied to the experimental specimens with a syringe into a half-circle channel
0.4 mm in diameter printed into the face of the specimens. As exemplified in Fig. S1(a), additional Ag-
TPU ink is injected into a small portion of the 1.4 mm internal channels to help ensure a strong connection
at the location where the wire lead and ink meet. Epoxy (Hampton Research HR4-347) is injected into the
end of the 1.4 mm internal channels where the wires leave the specimen in order to prevent movement
between the wires and ink. Both the Ag-TPU ink and epoxy are allowed to cure at room temperature for 24
hours before experiments.
Overall dimensions for the strain-sensitive and strain-insensitive specimens are given in Figs. S1(b) and
(c). As shown in Fig. S1(b), the Ag-TPU ink is applied on the vertical 2.2 mm beam of the strain-sensitive
specimen. As shown in Fig. S2(c), the Ag-TPU ink is applied on the diagonal 1.8 mm member of the strain-
insensitive specimen. Due to the soft nature of the Tangoblackplus material and the need to stabilize the
prints during 3D printing, 1.8 mm plates made of Veroblackplus (RGD875) are printed on the top and
bottom of each specimen. With the differences in Shore hardness values (Shore 26A and Shore 83D for
Tangoblackplus and Veroblackplus, respectively), it is assumed that the thin Veroblackplus plates act as
2
rigid bodies with respect to the rest of the Tangoblackplus specimen. Therefore, strains for the strain-
sensitive and strain-insensitive specimens are calculated with respect to heights of 22.9 mm and 22.1 mm,
respectively. These dimensions correspond to the portion of the heights of each specimen composed of
Tangoblackplus as seen in Fig. S1(b) and (c).
Figure S1. (a) Schematic of specimen fabrication constituents, shown specifically for the strain-sensitive specimen. Overall
dimensions are shown for (b) the strain-sensitive specimen and (c) strain-insensitive specimen.
2 Viscoelastic hyperelastic material model parameter identification
To characterize the frequency and time dependent influences observed experimentally, a Prony series,
viscoelastic hyperelastic material model is utilized. A cylindrical specimen of the same base material as the
flexible hybrid electronic material system specimens undergoes DMA experiments in a mechanical
spectrometer (TA Instruments ARES Rheometer). The DMA data in Fig. S2 is used to empirically identify
the viscoelastic material model by the ABAQUS internal fitting routines for a Prony-series material model,
with computed parameters shown in Table 1. The Prony series contributes to time dependent material
behavior in accordance with (1) and (2). The ig and the
ik are respectively the thi shear and bulk relaxation
coefficients, while i is the thi time relaxation constant. The constants 0
10 / 2C and 0
1 2 /D are
respectively the instantaneous material properties of the hyperelastic material model.
/0
10 10
1
1 1 j
NtR
j
j
C t C g e
(1)
/0
1 1
1
/ 1 1 j
NtR
j
j
D t D k e
(2)
3
Figure S2. DMA experimental data from 3 to 50 Hz conducted on a Tangoblackplus cylindrical specimen used in ABAQUS
FE models to generate the viscoelastic material model.
Table 1. Prony-series parameters computed by ABAQUS to generate the viscoelastic material model
i ig ik i
1 0.79215 0 0.00175
2 0.1202 0 0.0251
3 0 0.80177 0.00164
4 0 0.11483 0.0256
3 FE model studies for instantaneously applied stress in conductive ink traces
It is reported that instantaneously applied stress may result in a deterioration of conductive networks where
larger stresses are associated with larger initial increases in resistance [1] [2]. This influence may contribute
to the findings in Sec. 3.1 of the main text that show a 35% and 8% change in resistance by displacement
cycle 100 for the =4% and 8% configurations of the strain-sensitive specimen, as shown in Fig. 3(c).
Here, FE models of the specimens are used to characterize changes in von Mises stress experienced by the
conductive ink traces at the beginning of a cyclic displacement experiment at 25 Hz. In order to model the
experimental conditions, a downward displacement of 0.125 mm is applied to the specimens over the course
of 0.02 seconds in a dynamic, explicit FE model formulation. The value of downward displacement 0.125
mm is the amount of downward displacement experienced by the specimens for half of a displacement
cycle at 25 Hz, which takes 0.02 seconds to complete. Von Mises stress within the center of the beams onto
which the conductive ink trace is applied, as shown in Fig. S1(b) and (c), is then monitored for the specimen
geometries.
For the cyclic displacement experiments of Sec. 3.1 in the main text, the resistance change trends of Fig.
3(c) of the strain-sensitive specimen suggest that a rapid initial change in resistance occurs by displacement
4
cycle 100 for the =4% and 8% configurations. The FE model results in Figures S3(a) for the strain-
sensitive specimen show that the =4% and 8% configurations are associated with changes in peak stress
of 13100 Pa and 8500 Pa, respectively, after the application of the 0.125 mm displacement. On the other
hand, the peak stress change of the =14% configuration is 3300 Pa. The resistances measured for the
=14% strain-sensitive specimen in Fig. 3(c) do not indicate significant initial resistance increase and suggest
that the resistance change of 1% by displacement cycle 100 may be more associated with applied cyclic
displacements than with initial peak stress changes. The changes in peak stress for the =4% and 8% strain-
sensitive specimen configurations in Fig. S3(a) represent approximately 400% and 260%, respectively, the
change in peak stress experienced by the =14% configuration. The corresponding resistance changes of
the =4% and 8% strain-sensitive specimen configurations at displacement cycle 100 are approximately
35% and 8% the values measured for the =14% configuration. Therefore, there may be underlying
relations between instantaneous stress change and initial resistance change for the strain-sensitive FHE
material system geometry.
For the strain-insensitive specimen, Fig. S3(b) shows peak stress changes of 1500 Pa, 1300 Pa, and 330 Pa
for the =4%, 8%, and 14% configurations, respectively. The corresponding resistance change
measurements in Fig. 3(d) do not indicate a large increase in resistance at the beginning of the experiments,
and thus do not suggest that peak stress changes within the conductive ink traces of the strain-insensitive
specimen contribute greatly to resistance changes.
The FE model findings of peak stress changes within the conductive ink traces suggest that they may
contribute to the initial resistance changes shown in Figs. 3(c) and (d) of the main text such that larger
changes in peak stress, Fig. S3, are associated with larger initial changes in resistance, the 100-cycle data
in Fig. 3(c,d). Yet, future research should explore how specimen geometry and configuration contribute to
resistance changes within conductive networks due to these initial transient sequences of dynamic load
events.
Figure S3. Change in ink trace stress for excitation applied at beginning of 25 Hz experiment for =4%, 8%, and 14%
configurations of (a) strain-sensitive and (b) strain-insensitive specimens
5
4 Visual disruption of conductive ink network due to applied strain
For demonstrative purposes, 20% strain is applied to a Tangoblackplus substrate with a conductive ink trace
on the center. The substrate is 41.5 mm x 17.5 mm x 2.6 mm in overall dimension, while the conductive
ink trace is 27.2 mm x 2.5 mm in overall dimension and is aligned and centered on the substrate. Figures
S4(a) and (b) show the conductive ink network before and after the applied strain, respectively, with inset
schematics of the specimen included for reference. Disruptions to conductive ink trace networks, as shown
in Fig. S4(b), change electrical resistance behavior and are noticeable as distortions to the original network
in Fig. S4(a).
Figure S4. Visible disruption of a silver microflake network after 20% applied strain. Scale bars are 50 um.
5 Cyclic displacement experiments and modeling for specimens without conductive ink traces
As shown in Section 3.1 of the main text, the force transfer function of the specimens increases by less than
0.8% for all specimens and configurations by displacement cycle 10,000 in cyclic displacement
experiments. In Section 3.2 of the main text, it is hypothesized that increases in force transfer function occur
due to a softening of the elastomer material substrate of the specimen geometry. The softening of the
substrate corresponds to a greater softening of input force compared to output force because the input side
of the specimen is displaced whereas the outside side of the specimen is fixed.
In order to explore the influence that material softening may have on the force transfer function of the
specimens and to eliminate the influence of the heat generated by the conductive ink trace, cyclic
displacement experiments are performed on identical specimens without conductive ink traces. The
experiments are conducted with 0.25 mm peak-peak applied displacement for 10,000 displacement cycles.
The force transfer function is shown for the strain-sensitive and strain-insensitive specimens in Figures
S5(a) and (b), respectively. Additionally, using the FE model described in Sec. 2.2 of the main text, the
cyclic displacement experiments are modeled for the =4% configurations of the strain-sensitive and
6
strain-insensitive specimens for comparison. Note that the FE model described in Sec. 2.2 of the main text
does not account for the thermoelasticity of the substrate material.
As shown in the experimental results of Figs. S5(a) and (b), the force transfer function increases for the
strain-sensitive and strain-insensitive specimens by less than 0.8% for all configurations. The results in
Figs. S5(a) and (b) resemble those in Figs. 3(a) and (b) of the main text for the specimens with conductive
ink traces. Therefore, without the influence of the heat generated by the conductive ink trace, the results in
Figs. S5(a) and (b) suggest that the changes in force transfer function measured in Figs. 3(a) and (b) are due
to a softening of the elastomer material that makes up the specimen geometry, for which the displaced
excited input side of the specimens softens more than the fixed output side. Furthermore, the FE model
simulation results for the =4% configurations of both specimens in Figs. S5(a) and (b) shows a constant
force transfer function with number of displacement cycles. The force transfer function computed in the FE
model indicates that the changes in force transfer function measured experimentally for both specimens are
due to internal heat generation of the elastomer material substrates, since the FE model does not account
for these influences. The results shown in Fig. S5 conclusively identify the primary origins of the softening
and heat generation of the specimen to be associated with the thermoelasticity of the substrate material
instead of being associated with minor contributions from the Ag-TPU ink trace.
Figure S5. Force transfer function for cyclic displacement experiments conducted on specimens without conductive ink
traces is shown for =4%, 8%, and 14% configurations of the (a) strain-sensitive and (b) strain-insensitive specimens for
a 10,000 cycle displacement experiment with 0.25 mm peak-to-peak applied displacement. FE model simulation results for
the cyclic displacement experiments are shown for the =4% configuration of both specimens as well.
6 FE model calculations for conductive ink trace strain rate
It is shown that cumulative damage in conductive networks increases with increasing strain [3] [4] and
strain rate [5] [6] over the duration of cyclic displacement loading. These trends are thought to be due to
stress propagation in the conductive networks. In order to investigate the potential of these influences on
the conductive ink traces in this research, FE models of the specimens, as detailed in the Sec. 2.2 of the
main text, are used to calculate the strain rate that the ink trace experiences under the various geometries
7
and configurations in this research. The strain rate of the conductive ink trace beam for each specimen and
configuration is calculated for a 0.25 mm peak-to-peak harmonic displacement applied to the FE models of
the specimens at displacement frequencies ranging from 3 to 50 Hz. In order to calculate the strain rate of
the conductive ink traces, the peak-to-peak length changes of the conductive ink traces for each excitation
frequency are calculated in the FE model and are normalized with respect to the length of the conductive
ink trace under static conditions for each configuration. As shown in Figures S6(a) and (b) for the strain-
sensitive and strain-insensitive specimens, respectively, FE model results show that the strain rate that the
conductive ink trace experiences decreases with increasing specimen compression for both specimens.
Additionally, strain rate is shown to increase with displacement frequency for all specimens and
configurations. The FE model results presented here may have implications for the frequency dependence
of resistance for the experimental specimens, such that increasing strain rates and displacement excitation
frequency may result in increases in resistance.
Figure S6. Strain rate of conductive ink trace according to excitation frequency and configuration for the (a) strain-sensitive
and (b) strain-insensitive specimens
References
[1] J.T. Muth, D.M. Vogt, R.L. Truby, Y. Menguc, D.B. Kolesky, R.J. Wood, and J.A. Lewis, Embedded
3D printing of strain sensors within highly stretchable elastomers. Advanced Materials 26 (2014) 6307-
6312.
[2] L. Wang, T. Ding, and P. Wang, Effects of instantaneous compression pressure on electrical resistance
of carbon black filled silicone rubber composite during compressive stress relaxation. Composites
Science and Technology 68 (2008) 3448-3450.
[3] A. Jahanshahi, P. Salvo, and J. Vanfleteren, Reliable strechable gold interconnects in biocompatible
elastomers. Journal of Polymer Science 50 (2012) 773-776.
8
[4] S. Wagner, S.P. Lacour, J. Jones, P.I. Hsu, J.C. Sturm, T. Li, and Z. Suo, Electronic skin: architecture
and components. Physica E: Low-dimensional Systems and Nanostructures 25 (2004) 326-334.
[5] M. Borghetti, M. Serpelloni, E. Sardini, and S. Pandini, Mechanical behavior of strain based on
PEDOT:PSS and silver nanoparticles inks deposited on polymer subtrate by inkjet printing. Sensors
and Actuators A: Physical 243 (2016) 71-80.
[6] N.C. Das, T.K. Chaki, and D. Khastgir, Effect of axial stretching on electrical resistivity of short carbon
fibre and carbon black filled conductive rubber composites. Polymer International 51 (2002) 156-163.